EKF: Convert GPS yaw observation method to use SymPy generated code

[priseborough]

This PR replaces the Matlab generated equations with sympy generated equations. Continues on from #870.

Modifies the methods used for fusion of GPS receiver yaw angle.
Includes comparison test scripts that compare the observation Jacobians and Kalman gain equations generated for the Matlab symbolic toolbox and SymPy.

Testing

gps_yaw_fusion_generated_compare.cpp output:

Pass: GPS yaw Hfusion max diff fraction = 1.796981e-07
Pass: GPS yaw Kfusion max diff fraction = 3.981467e-07

SITL testing TODO

[priseborough]

If someone with a RTK yaw GPs setup can provide a log for replay, then that would be appreciated. TXS.

[bresch]

@priseborough This is a log that I used in the past to replay GPS yaw data: https://logs.px4.io/plot_app?log=46feda49-d644-4ed0-acbd-868ff70779eb

[priseborough]

I was able to test this in SITL using a modification to ekf2_main.cpp that set the gps.heading value using the vehicle_attitude_ground_truth message and EKF2_AID_MASK = 129

A replay comparison using the log provided gave identical innovation and innovation variance sequences before:

and after the change:

[priseborough]

@bresch and @kamilritz My understanding is that the review issues have been addressed and testing is compete.

[kamilritz]

@priseborough I would prefer, when we do the mag one first and copy the new standard way of handling the measurement jacobian to here.

[priseborough]

OK, can wait for you to get the matrix library changes required into firmware.

[priseborough]

Here are the before at commit and after at commit 8e52aed results which are visually identical.

Before:

After:

[kamilritz]

I just added the missing file with the auto generated code. Is there a reason why we now do not write ecl::powf?

[priseborough]

I just added the missing file with the auto generated code. Is there a reason why we now do not write ecl::powf?

We could incorporate a text find and replace, but it would have to check that the exponent was an integer.

Changes have been incorporated into updates and approved by second reviewer